An Algorithm for Fabric Defect Detection Based on Adaptive Canny Operator
Authors:
- 1 December 2019
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 475-481
Abstract
Aiming at the problem that the parameters of traditional Canny operator need to be set manually and the adaptability of the algorithm is poor, this paper proposes an improved Canny operator algorithm for weaving damage detection. Firstly, the detection principle of traditional Canny operator is analyzed; secondly, the image processing of knitted fabric is improved by iterative filtering and single-scale Retinex (Retina and Cortex abbreviation) non-linear image enhancement method; then, the high and low thresholds of Canny operator are adaptively set by the method of maximum inter-class variance; finally, the algorithm is obtained by field experiments and the accuracy rate is 0.97. Compared with the traditional Canny operator, it has better adaptability and detection accuracy. At the same time, it can describe the details of knitted fabric defects better.Keywords
This publication has 6 references indexed in Scilit:
- Automatic Defect Segmentation for Plain Woven Fabric Images2019
- Prewitt edge detection based on BM3D image denoising2018
- Automated textile defect classification by Bayesian classifier based on statistical features2017
- Combined Feature Extraction and Selection in Texture Analysis2016
- Fabric defect detection systems and methods—A systematic literature reviewOptik, 2016
- Fabric Defects' Automatic Inspection Based on Computer Vision2009